Freelancer profile translated to English.

Description

What I like about data is taking a vague business problem and turning it into something that actually runs in production, not just a beautiful notebook.

For 3 years, I've been navigating between pipelines (Kafka, Airflow, n8n), ML/DL/NLP models, and increasingly, integrating LLMs into concrete workflows.

I enjoy coding as much as understanding why a business team really needs one figure over another; that's often where the difference between a useful data project and a data project forgotten in a corner is made.

Languages

Arabic
Native or bilingual
French
Native or bilingual
English
Fluent

Workplace preferences

Can work on-site

Paris (up to 50km)

Drivecase / Vanberg (Ymajor)
Data Consultant
April 2026 - June 2026 (2 months)
Strasbourg, France
• • Structuring of ESG KPIs for EcoVadis certification (environmental, social, ethical): definition of indicators, performance monitoring, preparation and organization of compliance documents - direct contribution to the score's progression, from Bronze to Platinum medal.
• • Analysis of stock discrepancies between theoretical and physical data using Excel, SQL, and Zoho Analytics: identification of item references causing anomalies and quantification of associated financial losses, enabling actionable prioritization of corrections by business teams. (Drivecase)
• • Design and deployment of an invoice processing pipeline with n8n: automated ingestion, OCR extraction via Mistral AI (factur-X invoices) and XML parsing (structured invoices) - addition of an intermediate layer for data control and observation, replacing a direct unsupervised flow to the data warehouse, normalization, and loading into Zoho Books. (Drivecase)
Environment: Python, SQL, Excel, n8n, OCR (Mistral AI), Zoho Analytics, Zoho Books, JavaScript
Python SQL Microsoft Excel n8n Zoho Books
CMA CGM
Full Stack Data Engineer
July 2024 - July 2025 (1 year)
Marseille, France
• • Development and deployment of a Random Forest model (Regression, 99.99% accuracy) to compensate for GPS signal loss on terminal equipment.
• • Modeling and transformation of heterogeneous data (CSV, JSON, APIs): cleaning, structuring, and creation of complex DAX measures for Power BI dashboards used by business teams.
• • Design and deployment of a real-time monitoring system for maritime terminals: implementation of an intermediate layer (Kafka, Docker containerization) ensuring the recovery, cleaning, and normalization of raw data from connected equipment before re-injection into the data warehouse - significant reduction in anomaly detection time. Kibana visualization.
• • Management of critical data persistence via PostgreSQL and Redis: ensuring integrity, high availability, and low latency.
Apache Kafka Python Node.js Microsoft Power BI Data visualisation
ONTBO
Data Engineer – NLP & Emotion Detection
March 2024 - September 2024 (6 months)
60200 Compiègne, France
• • Migration of the emotion detection system from the Ekman model (7 basic emotions) to a 3-dimensional vector model (Valence, Arousal, Dominance): the intersection of these axes allows for the detection of over 30 nuanced emotions, compared to 7 initially - significant gain in granularity for sentiment analysis.
• • Fine-tuning and optimization of pre-trained models via Reinforcement Learning from Human Feedback: continuous improvement of the F1-score by +12% on minority classes.
• • Construction of an end-to-end pipeline: annotated data collection → cleaning → training → evaluation → production performance monitoring.
Python Deep Learning Feature Engineering